With the dramatic growth of the image data on the web, there is an increasing demand of the algorithms capable of understanding the visual information automatically.

The invention of neural networks marks a critical milestone in the pursuit of true artificial intelligence. Despite their impressive performance on various tasks, these networks face limitations in learning efficiently as they are often trained from scratch.

This thesis mainly focuses on multimodal understanding and Visual Question Answering (VQA) via deep learning methods. For technical contributions, this thesis first focuses on improving multimodal fusion schemes via multi-stage vision-language interactions.

This thesis mainly focuses on cross-modal retrieval and single-modal image retrieval via deep learning methods, i.e. by using deep convolutional neural networks.

In 2018, the number of mobile phone users will reach about 4.9 billion. Assuming an average of 5 photos taken per day using the built-in cameras would result in about 9 trillion photos annually.

In this thesis, deep learning is studied from a statistical perspective. Convergence rates for the worst case risk bounds of neural network estimators are obtained in the classification, density estimation and linear regression model.

Tomography is a powerful technique to non-destructively determine the interior structure of an object.Usually, a series of projection images (e.g.\ X-ray images) is acquired from a range of different positions.

Image segmentation is a fundamental task in computer vision, with applications ranging from medical diagnostics to archaeological research.

We are living in an information era where the amount of image and video data increases exponentially.

The recently introduced nature-originated deep eutectic solvents, so-called natural deep eutectic solvents (NADES) are considered as truly green solvents, which composed of natural ingredients found abundantly in organisms.

Op dinsdag 26 juni 2018 verdedigde Linlin Sun haar proefschrift ‘International Environmental Obligations and Liabilities in Deep Seabed Mining’. Het onderzoek is begeleid door promotoren prof. dr. N.J. Schrijver en prof. dr. E.C.P.D.C. De Brabandere.

In deep reinforcement learning, searching and learning techniques are two important components. They can be used independently and in combination to deal with different problems in AI, and have achieved impressive results in game playing and robotics. These results have inspired research into artificial…

With the emergence of online social networks (OSNs), the way people create and share information has changed, which becomes faster and broader than traditional social media.

Building a Deep Cross-media Transfer Network to extract visual relations that relieve the problem of insufficient training data for visual tasks.

A brief review essay on some of the work that has been recently published in the emergent field of digital ethnography.

This thesis aims to enhance our understanding of galaxies by testing theoretical models of galaxy formation against observations, particularly in the cases of extreme systems which have been found to have an excess of baryonic mass in their central regions, in the form of either supermassive black holes…

Op dinsdag 26 juni 2018 verdedigt Linlin Sun haar proefschrift ‘International Environmental Obligations and Liabilities in Deep Seabed Mining’. De promotoren zijn prof. dr. N.J. Schrijver en prof. dr. E.C.P.D.C. De Brabandere.

In this thesis we explore machine and deep learning approaches that address keychallenges in high dimensional problem areas and also in improving accuracy in wellknown problems. In high dimensional contexts, we have focused on computational fluid dynamics (CFD) simulations.

De theoretische wereld van wiskunde en informatica naar meer toegepaste onderzoeksgebieden brengen. Dat is wat Daan Pelt, universitair docent bij het Leiden Institute of Advanced Computer Science (LIACS) probeert te bereiken met de methodes die hij ontwikkelt als oplossing voor uitdagingen in beeldv…

This thesis investigates the application of deep learning techniques in image analysis across various domains, focusing on four main themes: feature extraction, classification, segmentation, and integration, demonstrating the transformative potential of these technologies.

In this thesis, we have studied cell images from two types of cells, including pollen grains and the immune cells, neutrophils. These images are captured using a bright field microscope and a confocal microscope.

Radiography is an important technique to inspect objects, with applications in airports and hospitals. X-ray imaging is also essential in industry, for instance in food safety checks for the presence of foreign objects.

Over several decades, a variety of computational methods for drug discovery have been proposed and applied in practice. With the accumulation of data and the development of machine learning methods, computational drug design methods have gradually shifted to a new paradigm, i.e. deep learning methods…

Sewer pipes are an essential infrastructure in modern society and their proper operation is important for public health. To keep sewer pipes operational as much as possible, periodical inspections for defects are performed.

In this thesis, we use low-frequency and high-frequency radio observations to address the following questions regarding quasars: is the radio loud/quiet quasar dichotomy real?

Een intelligente vertaalmachine gebaseerd op nieuwe ontwikkelingen in machine learning gaat de concurrentie aan met grote vertaalservices als die van Google. Maar ook Google maakt grote vorderingen sinds het is het is overgestapt op deep learning, aldus informatica-onderzoeker Suzan Verberne in de V…

In this thesis, topics relating to the optimization of high-throughput pipelines used for imaging are discussed. In particular, different levels of implementation, i.e., conceptual, software, and hardware, are discussed and the thesis outlines how advances on each level need to be made to make gains…

Een computer kan veel meer beelden bekijken, en er van leren, dan een menselijke specialist. Voor medische en biologische toepassingen worden AI-systemen in hoog tempo onmisbaar. Ze moeten nog wel leren uitleggen hoe ze tot hun oordeel komen.

This thesis utilises the exceptional sensitivity and resolution of LOFAR to explore the low-frequency Universe, while simultaneously advancing the telescope's capabilities to enable new scientific discovery.

We study the interplay of topology and geometry with chirality for several passive and active systems, employing both analytical and numerical methods.

Hoe is Google’s computerprogramma AlphaGo zo goed geworden in bordspel Go dat het de wereldkampioen kon verslaan? Op 29 juni vertelt ontwikkelaar Aja Huang daarover tijdens een lezing in het Gorlaeusgebouw.

De voorspelling van Jaap van den Herik, directeur van het Leiden Centre of Data Science, is uitgekomen: Google-computer AlpaGo heeft de wereldkampioen Go verslagen met 4-1. In interviews met de Volkskrant en BNR Nieuwsradio vertelt Van den Herik over het toernooi én over de volgende uitdaging: het spel…

Hoe is Google’s computerprogramma AlphaGo zo krachtig geworden? Op 29 juni vertelde ontwikkelaar Aja Huang daarover tijdens een lezing in het Gorlaeusgebouw.

Lezing

Je kunt aan de Universiteit Leiden Kunstmatige Intelligentie studeren, maar ook binnen andere opleidingen speelt het een steeds grotere rol.

De Faculteit Archeologie

Promotor: Prof.dr. M.L. van Heck, Co-Promotor: B.P. Tighe

In this course, you will gain a deep insight into the rules that govern armed conflict, and aim to mitigate human suffering on the battlefield. You will explore the why and how of International Humanitarian Law, followed by the different types of conflict.

Het Computer Systems, Imagery & Media (CSI) onderzoeksprogramma voert fundamenteel onderzoek uit naar methoden en technieken voor het ontwerp, de implementatie en de applicatie van geavanceerde computersystemen.

FAIR-ASSESS wordt bijgestaan door een wetenschappelijke raad bestaande uit interne leden verbonden aan Universiteit Leiden en externe leden van buiten de universiteit met expertise in het veld.

Computers zijn in staat om ongelooflijk nauwkeurige voorspellingen te doen op basis van machine learning. Met andere woorden, deze computers kunnen zonder tussenkomst leren als ze eenmaal door mensen zijn voorgeprogrammeerd. Bij LIACS verkennen en verleggen we de grenzen van wat een revolutionaire nieuwe…